An 8-Element Frequency-Selective Acoustic Beamformer and Bitstream Feature Extractor with 60 Mel-Frequency Energy Features Enabling 95% Speech Recognition Accuracy.
Seungjong LeeTaewook KangJohn BellMohammad R. HaghighatAlberto J. MartinezMichael P. FlynnPublished in: VLSI Circuits (2020)
Keyphrases
- recognition accuracy
- feature extractor
- bitstream
- feature extraction
- face recognition
- recognition rate
- bit rate
- coding scheme
- feature space
- feature vectors
- feature set
- multiresolution
- feature values
- multipath
- error resilience
- wavelet transform
- speech recognition
- wavelet coefficients
- end to end
- multiscale
- high quality
- video quality
- error concealment
- bit plane